Data for the tropical upper troposphere (8-12 km, 20° N-20° S) collected during NASA's Pacific Exploratory Missions have been used to carry out a detailed examination of the photochemical processes controlling HOx (OH+HO2). Of particular significance is the availability of measurements of nonmethane hydrocarbons, oxygenated hydrocarbons (i.e., acetone, methanol, and ethanol) and peroxides (i.e., H2O2 and CH3OOH). These observations have provided constraints on model calculations permitting an assessment of the potential impact of these species on the levels of HOx, CH3O2, CH2O, as well as ozone budget parameters. Sensitivity calculations using a time-dependent photochemical box model show that when constrained by measured values of the above oxygenated species, model estimated HOx levels are elevated relative to unconstrained calculations. The impact of constraining these species was found to increase with altitude, reflecting the systematic roll-off in water vapor mixing ratios with altitude. At 11-12 km, overall increases in HOx approached a factor of 2 with somewhat larger increases being found for gross and net photochemical production of ozone. While significant, the impact on HOx due to peroxides appears to be less than previously estimated. In particular, observations of elevated H2O2 levels may be more influenced by local photochemistry than by convective transport. Issues related to the uncertainty in high-altitude water vapor levels and the possibility of other contributing sources of HOx are discussed. Finally, it is noted that the uncertainties in gas kinetic rate coefficients at the low temperatures of the upper troposphere and as well as OH sensor calibrations should be areas of continued investigation. Copyright 1999 by the American Geophysical Union.